Subscribe Bookmark RSS Feed

Adjustment for variable (age, weight, sex, BMI, etc.)

jxa014

Community Trekker

Joined:

Mar 30, 2016

I am running a one-way ANOVA and am looking for a solution to match/adjust the analysis based on my demographic variables (age, weight, etc.). 

 

Any solution to this?

Best, 

JD

1 ACCEPTED SOLUTION

Accepted Solutions
Dan_Obermiller

Joined:

Apr 3, 2013

Solution

The age, weight, etc. are usually considered covariates. So you usually will conduct the analysis by using Fit Model rather than Fit Y by X. You would add your primary predictor variable along with all of the other possible demographic variables into the model. There are some potential pitfalls that would be too lengthy to discuss in this forum. I would recommend looking at a linear models text. JMP also offers classes on fitting these types of models: https://support.sas.com/edu/schedules.html?ctry=us&crs=JANR

Dan Obermiller
7 REPLIES
Dan_Obermiller

Joined:

Apr 3, 2013

Solution

The age, weight, etc. are usually considered covariates. So you usually will conduct the analysis by using Fit Model rather than Fit Y by X. You would add your primary predictor variable along with all of the other possible demographic variables into the model. There are some potential pitfalls that would be too lengthy to discuss in this forum. I would recommend looking at a linear models text. JMP also offers classes on fitting these types of models: https://support.sas.com/edu/schedules.html?ctry=us&crs=JANR

Dan Obermiller
jxa014

Community Trekker

Joined:

Mar 30, 2016

Dan, 

Do you happen to have an output of that process? Or a step by step? Regardless of pitfalls, just trying to do a simple analysis adjusted for a few variables. 

Dan_Obermiller

Joined:

Apr 3, 2013

Look in the online JMP manual: Fitting Linear Models. Look on page 214 for the Analysis of Covariance with Unequal Slopes Example. This is not the same as your example, as your situation sounds more complex. But this will give you an idea of how to specify the model and what a small part of what the output may look like. The full output could be seen in just about any standard least squares multiple regression model that is fit in JMP.

 

Although your question can be simply stated, the analysis may not be so easy. If it were easy, there would be no need for the discussion here. There are many issues to consider such as: Is there collinearity in the data? Are the data complete or are there "gaps"? Are there time-related effects? Are the effects fixed or random? Should interactions be considered? The devil is always in the details. Much more information would be needed to be certain that proper advice is given. You may wish to refer to a text on linear models such as Applied Linear Statistical Models by Kutner, Neter, et. al.

Dan Obermiller
Ted

Community Trekker

Joined:

Mar 29, 2016

 I would start by defining the aim of the study and the target variable (endpoint).

jxa014

Community Trekker

Joined:

Mar 30, 2016

Ted, 

I am looking at a genotype that has 3 subtypes (1:1, 1:2, 2:2) and observing their relationship to glucagon. 

 

My genotype is nominal and my glucagon is continuous. 

 

I have several other continuous variables (covariates) such as: body weight, age, and gender. 

 

I am wanting to run an ANCOVA with glucagon as my response variable and my genotype as my factor. My covariates are as I said previously. 

 

Does that clear it up?

markbailey

Staff

Joined:

Jun 23, 2011

  1. Select Analyze > Fit Model.
  2. Select Genotype and click Y.
  3. Select Glucagon and all the covariates.
  4. Do you expect interaction effects?
    1. If so, click Macros and select Factorial to Degree (2).
    2. If not, click Add.
  5. Click Run.
Learn it once, use it forever!
Ted

Community Trekker

Joined:

Mar 29, 2016

I think, it's possible to use "Analyze > Fit Y by X", if the continuous variables to transform to discrete (better binary). For example, for age take less/more than 60, for weight a certain number that divides population into normal weight and obese, etc. Anyway: Select Glucagon as Y.